== Installed Camera Information

**Hikvision DS-2CD5585G0-IZHS** 
    Mudd floor 1, towards 120th street
    Mudd first floor Mech E lab
    [https://www.hikvision.com/uploadfile/image/11769_DatasheetofDS2CD5585G0IZ(H)S.pdf]

**Hikvision DS-2CD5585G0-IZHS**
    Mudd 2nd floor, towards Amsterdam balcony camera

**Hikvision DS-2CD4AC5F-IZH ** 
    12MP Outdoor Network Bullet Camera with Night Vision, 2.8-12mm Lens & Built-In Heater S/N: 212528616:
    20191015: installation in Botwinick lab mudd1224, towards 120th street 
**Hikvision DS-2CD5585G0-IZHS  ** 
    2.8mm-12mm (8MP)
    Mudd 12thFloor Balcony towards Amsterdam Avenue


== COSMOS Cameras Data-set
**1st-floor videos (anonymized*)**: [https://drive.google.com/drive/u/0/folders/1QXrfsLXEKfRfQyc6qzvtg37A0Z1i0io5] 

**2nd-floor videos (anonymized*)**: [https://drive.google.com/drive/u/0/folders/1LR7H4theRazz2_uYHvCFGVVewQmKbWSF]

**12th-floor videos (120th street)**: [https://drive.google.com/drive/u/0/folders/1SEsocAAIReepdjE4XyVyT4kiqrunv7BU]

**12th-floor videos (Amsterdam Avenue)**: [https://drive.google.com/drive/u/0/folders/1qC-62s8ohTGg-odyzo7BNw2GDv1OIeoK]

Videos in this directory are outputs of the COSMOS YOLOv4 blurring pipeline.

Faces and license plates are anonymized with Gaussian blurred areas defined by bounding box detection coordinates. A high-level overview of the blurring.

***Anonymization workflow:**

1. Frames are read individually from a video file.

2. Each frame is then:

    2.1 Resized to the input size of the specific YOLOv4 model (960x960 or 1440x1440)

    2.2 Passed through the YOLOv4 model which outputs bounding box predictions

    2.3 Blurred corresponding to the bounding box predictions

3. Blurred frames are written to the output video file.

The YOLOv4 blurring model is trained in Darknet on the Mudd 1st floor video
dataset annotated in Summer 2021. Models are converted from Darknet to PyTorch for
integration into the current implementation of the blurring pipeline.